Background: Smoking was classified as “possibly carcinogenic” (hazard classification 2B) for breast cancer in the 2012 International Agency for Research on Cancer Monograph. The re-evaluation, for both second-hand smoke and active smoking, is expected 2027-2029. Tobacco use is currently listed by the World Health Organization as one of several risk factors that increase the risk of breast cancer. The Premenopausal Breast Cancer Collaborative Group (PBCCG) is a rich resource for studying risk factors for premenopausal breast cancer, which may be different from postmenopausal breast cancer.
Objectives: To examine the association of key measures of active smoking with risk of premenopausal breast cancer.
Methods: We followed 718 462 women, who were enrolled at 16-54 years of age in 16 cohort studies from 1976-2012. The cohorts from North America, Europe, and Australia identified incident breast cancer cases over a median follow-up of 8.6 years. We used Cox proportional hazards regression to estimate study multivariable-adjusted hazard ratios (HRs) with 95% confidence intervals (CIs). We examined associations with ever compared with never smoking at cohort entry, age at smoking initiation (<15, 15-19, ≥20 years), smoking duration (≤10, 11–20, ≥21 years), number of pack-years (≤10, 11–20, ≥21), and years smoked before first childbirth (1-5,6-10,11+years). Potential confounding factors, identified a priori, included in models were education, body mass index, age at menarche, age at first childbirth, number of births, and alcohol consumption. Study-specific estimates were combined using random-effects meta-analysis. Not all cohorts were included in all analyses. Tests for linear trends were conducted by treating the categorical exposure variables as continuous in the models.
Results
Altogether 12 972 incident premenopausal breast cancer cases were identified. The proportion of ever smokers ranged from 24.3% in the California Teacher’s Study to 56.4% in the Southern Community Cohort Study. The multivariable adjusted HR for ever compared with never smokers was 1.03 (95% CI; 0.96-1.10). Compared with never smokers, the multivariable adjusted HR estimate for ever smokers who had initiated smoking before age 15 years was 1.12 (95% CI; 1.00 -1.25), while that for those who had smoked for more than 20 years and those who had smoked more than 20 packyears were 1.11  (95% CI; 1.01-;1.22), ) and  1.12 (95% CI; 1.02-1.23), respectively. Compared with parous never smokers, parous women who had started to smoke more than ten years before their first birth had a 15% higher risk (HR 1.15; 95% CI; 1.06-1.26) We observed a dose-response relationship with risk of premenopausal breast cancer for  smoking duration (p trend = 0.03) , number of pack years smoked (p trend <0.001), and number of years smoked before the first birth (p trend <0.0001), but not for age at smoking initiation (p trend = 0.09).
Conclusions/Implications Our findings from prospective studies across diverse populations suggest that cigarette smoking is associated with an increased risk of premenopausal breast cancer, adding to the evidence that premenopausal breast cancer should be considered a smoking-related cancer. The cancer burden due to smoking may be underestimated.